Engineering in the AI Age: How I Build Real Systems as a Sophomore

I'm a sophomore, which means I'm still early in my engineering career. I'm not going to pretend I'm a world-class programmer. I use AI tools heavily: sometimes for debugging, sometimes for boilerplate, sometimes for refactoring or filling knowledge gaps. That's the truth. But the part that actually matters is what I'm responsible for.

For every project I build, I'm responsible for the idea, the architecture, and the direction. I decide what the system is supposed to do, how data flows through it, how stages connect, and what the constraints are. AI accelerates the implementation, but it doesn't define the problem. It doesn't choose the tradeoffs. It doesn't set the boundaries. I do.

There's a misconception that "AI builds everything." It doesn't. If your idea is bad or your system design is broken, AI won't save you. It will produce more broken code, just faster. Good engineering still starts with understanding the problem. That part can't be automated.

I think of students like myself as people made of clay: we're moldable. The tools of our era shape how we learn. Before AI, students spent all their time memorizing data structures or grinding leetcode. Today, we integrate AI into our workflow because that's the environment we're growing up in. Being moldable isn't a weakness. It means we can adapt faster than older paradigms expect.

What I refuse to outsource to AI is the problem-finding process. If you can't walk through a city like New York and come up with five real problems to solve, you're not paying attention. Tools should amplify your curiosity, not replace it.

My workflow isn't a shortcut. It's alignment with the direction engineering is moving. But I still plan to strengthen my fundamentals. AI helps me build faster, but understanding systems deeply is what will make me a better engineer long-term.